Yes, the stuff applied/invented in industry has to work and generate a benifit. In academia I see it often that people stop after the proof of principle (which if the dataset is too small only appears to be a proof of principle)